Supplemental material to this article can be found at: http://dmd.aspetjournals.org/content/suppl/2015/10/06/dmd.115.065292.DC1

1521-009X/43/12/1934–1937$25.00 http://dx.doi.org/10.1124/dmd.115.065292 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 43:1934–1937, December 2015 Copyright ª 2015 by The American Society for Pharmacology and Experimental Therapeutics Short Communication

The Nonspecific Binding of Tyrosine Kinase Inhibitors to Human Liver Microsomes s

Received May 4, 2015; accepted October 5, 2015

ABSTRACT Drugs and other chemicals frequently bind nonspecifically to the shown to be unaffected by the addition of CHAPS (6 mM) to the constituents of an in vitro incubation mixture, particularly the dialysis medium. This approach was subsequently applied to mea- enzyme source [e.g., human liver microsomes (HLM)]. Correction for surement of the binding of axitinib, dabrafenib, erlotinib, gefitinib, nonspecific binding (NSB) is essential for the accurate calculation of ibrutinib, lapatinib, nilotinib, nintedanib, regorafenib, sorafenib, and Downloaded from the kinetic parameters Km, Clint, and Ki. Many tyrosine kinase trametinib to HLM (0.25 mg/ml). As with the validation set drugs, inhibitors (TKIs) are lipophilic organic bases that are nonionized at attainment of equilibrium was demonstrated in HLM-HLM and physiologic pH. Attempts to measure the NSB of several TKIs to HLM buffer-buffer control dialysis experiments. Values of the fraction by equilibrium dialysis proved unsuccessful, presumably due to the unbound to HLM ranged from 0.14 (regorafenib and sorafenib) to limited aqueous solubility of these compounds. Thus, the addition of 0.93 (nintedanib), and were generally consistent with the known

detergents to equilibrium dialysis samples was investigated as an physicochemical determinants of drug NSB. The extensive NSB of dmd.aspetjournals.org approach to measure the NSB of TKIs. The binding of six validation many TKIs to HLM underscores the importance of correction for TKI set nonionized lipophilic bases (felodipine, isradipine, loratidine, binding to HLM and, presumably, other enzyme sources present in in midazolam, nifedipine, and pazopanib) to HLM (0.25 mg/ml) was vitro incubation mixtures.

Introduction of many TKIs. However, NSB appears to have been accounted for in the at ASPET Journals on September 28, 2021 In vitro approaches, using human liver microsomes (HLMs), human calculation of kinetic parameters in just one report (Kenny et al., 2012). hepatocytes, or recombinant proteins as the enzyme source are used widely Given the increasing number of TKIs entering clinical practice, we to characterize the kinetics of drug metabolism and drug-drug interaction sought to characterize the drug-drug interaction potential of several potential (Houston, 1994; Obach, 1999; Rostami-Hodjegan and Tucker, drugs from this class. However, using lapatinib, pazopanib, regorafenib, 2007; Miners et al., 2010). However, numerous drugs, especially and sorafenib as the model compounds, we encountered major problems lipophilic organic bases and neutral compounds, bind nonspecifically to determining NSB by equilibrium dialysis with HLM as the enzyme the microsomal membrane. Importantly, nonspecific binding (NSB) source. In particular, equilibrium was established in buffer-buffer and reduces the concentration of the unbound drug in the incubation medium, HLM-HLM controls for dialysis times ranging from 3 to 24 hours only with pazopanib, irrespective of whether we used conventional equilib- resulting in overestimation of the measured kinetic constants Km (or S50) rium dialysis (employing dialysis cells) or a commercial rapid equilib- and the inhibitor constant Ki. Consequently, in vitro intrinsic clearance rium dialysis (RED) device. It has been reported that axitinib forms a (determined as Vmax/Km) and inhibitory drug-drug interaction potential supersaturated solution in buffers containing 0.5 or 1% dimethylsulfoxide (assessed as [I]/Ki,where[I] is the inhibitor concentration) are both underpredicted (Obach, 1999; McLure et al., 2000; Margolis and Obach, (DMSO) (Reyner et al., 2013), and we surmised a similar situation may 2003; Grime and Riley, 2006; Miners et al., 2006, 2010). Thus, it is now occur more generally with TKIs during the course of equilibrium dialysis widely accepted that NSB should be accounted for when kinetic experiments. Thus, we explored the use of a number of detergents parameters are calculated from in vitro metabolism and inhibition studies. to retain TKIs in solution without disrupting microsomal binding. As Tyrosine kinase inhibitors (TKIs) are an emerging group of drugs used reported here, inclusion of CHAPS in the dialysis medium allows in the treatment of many forms of cancer and some other diseases (e.g., determination of the NSB of weakly basic compounds, including TKIs. idiopathic pulmonary fibrosis). Typically, TKIs are lipophilic (log P values . 2.5) weak organic bases, which are frequently nonionized at physiologic Materials and Methods pH. Thus, it might be expected that TKIs exhibit significant NSB. There Materials. Axitinib, erlotinib, gefitinib, lapatinib, nilotinib, nintedanib, have been numerous reports of the in vitro enzyme and inhibition kinetics pazopanib, regorafenib, sorafenib, and trametinib were purchased from LC Laboratories (Woburn, MA); dabrafenib, ibrutinib, and isradipine were purchased from SelleckChem (Boston, MA); amitriptyline hydrochloride, Brij-58, CHAPS This study was supported by a project grant [1044063] from the National Health hydrate, chlorpromazine hydrochloride, felodipine, nifedipine, Triton X-100, and and Medical Research Council of Australia. zidovudine were purchased from Sigma-Aldrich (Sydney, Australia); and loratidine dx.doi.org/10.1124/dmd.115.065292. and midazolam were purchased from Toronto Research Chemicals (Toronto, ON, s This article has supplemental material available at dmd.aspetjournals.org. Canada). Solvents and other reagents were of analytical reagent grade. HLMs,

ABBREVIATIONS: CMC, critical micelle concentration; DMSO, dimethylsulfoxide; fumic, fraction unbound in incubation medium; HLM, human liver microsomes; HPLC, high-performance liquid chromatography; Ki, inhibitor constant; Km, kinetic constant; NSB, nonspecific binding; PB, phosphate buffer; RED, rapid equilibrium dialysis; TKI, tyrosine kinase inhibitor.

1934 The Nonspecific Binding of Tyrosine Kinase Inhibitors to HLM 1935 prepared according to the method of Bowalgaha et al. (2005), were sourced from High-Performance Liquid Chromatography. Concentrations of each drug the human liver “bank” of the Department of Clinical Pharmacology of Flinders present in the buffer and HLM compartments of dialysis cells (and RED devices) University. Approval was obtained from the Southern Adelaide Clinical Research were measured by HPLC. For all drugs except amitriptyline, axitinib, chlorprom- Ethics Committee for the use of human liver tissue in drug disposition studies in vitro. azine, nintedanib, and zidovudine, a 0.2-ml aliquot of sample from each cell of the Equilibrium Dialysis. The NSB of drugs to HLM was measured by equilibrium dialysis apparatus was treated with an equal volume of ice cold 4% acetic acid in dialysis. The equilibrium dialysis apparatus (Dianorm, Munich, Germany) com- methanol (for compatibility with the mobile phase composition), vortex mixed, and prised of Teflon dialysis cells of 1.2-ml capacity per side, separated by a Spectrapor then centrifuged at 5000g for 10 minutes. For the five drugs specified above, #4 dialysis membrane (molecular mass cut off 12,000–14,000 Da; Spectrum Medical samples were treated with perchloric acid (70%, 2 ml). A 20–60 ml aliquot of the Industries Inc., Los Angeles, CA). The dialysis membrane was soaked overnight in supernatant fraction was injected into the HPLC (Agilent 1100 series instrument; phosphate buffer (PB) (0.1 M, pH 7.4) at 4C prior to use in dialysis experiments. PB, Agilent Technologies, Sydney, Australia), which was fitted with a Waters a validation set or test drug (see Results and Discussion) dissolved in DMSO (0.01 NovaPak C18 analytical column (150 Â 3.9 mm (i.d.), 5-mm particle size; ml; 1% v/v) and HLM (0.25 mg/ml), was loaded into one side of each dialysis cell, Waters Corporation, Milford, MA). Chromatography conditions are shown in giving a total volume of 1 ml. Each of the drugs was diluted 1:100 upon addition to Supplemental Table 1. Chromatograms of all analytes were free from interference the dialysis cell to give the desired concentration. The other side of the dialysis cell under the conditions employed. Concentrations of validation set compounds and contained PB and DMSO (1% v/v). Dialysis experiments were performed in the TKIs were determined by reference to standard curves prepared for each compound presence and absence of detergent (see Results and Discussion) in each side of the over the ranges of 1–100 mM (non-TKI validation set compounds) and 0.1–2 mM dialysis cells (i.e., the same concentration of detergent was added to the donor and (TKIs). The fraction unbound to HLM (fumic) was calculated as the concentration of receiver compartments). The dialysis cell assembly was immersed in a water bath drug in the buffer cell divided by the concentration of drug in the cell containing – maintained at 37 C and rotated at 12 rpm. Equilibrium was achieved within 4 8 HLM. Downloaded from hours, although experiments were typically performed overnight. After dialysis, each Computational Calculation of Log P, Percent Ionization, and fumic. Log side of the cell was unloaded by expelling the contents into a 5-ml glass tube. Control P values were obtained using four algorithms. SMILES strings of compounds experiments, with either HLM (HLM-HLM) or PB (PB-PB) in both sides of the were obtained from the Pubchem server (https://pubchem.ncbi.nlm.nih.gov/) equilibrium dialysis cells (with or without detergent), were employed in each dialysis and then used to generate the two-dimensional structures of molecules in experiment to confirm attainment of equilibrium. The control experiments were MarvinSketch. The two-dimensional structures were used for calculating log performed at the lowest and highest drug concentration investigated (see Table 1). P values from Marvin (Marvin 14.12.8.0, ChemAxon, http://www.chemaxon.

Attainment of equilibrium in HLM-HLM and PB-PB control experiments was com; physico-chemical property predictors of the Marvin calculator plugin), dmd.aspetjournals.org deemed to occur for fraction unbound (fumic; see below) values in the range of 0.85– whereas AlogP2.1 (http://www.vcclab.org/web/alogps/) and XlogP3 (http://www. 1.05. Equilibrium dialysis experiments at each drug (validation and test sets) sioc-ccbg.ac.cn/software/xlogp3/) values were calculated from the respective concentration were performed in duplicate. Variances in duplicate fumic values were webservers using SMILES strings as the input. Log P values were additionally invariably ,5%. The NSB of lapatinib, pazopanib, regorafenib, and sorafenib to obtained from ACD/Laboratories ChemSpider (http://www.chemspider.com/) by a HLM (0.25 mg/ml) was also investigated using a commercial RED device (Thermo chemical name search. The pKa of each molecule was calculated at pH 7.4 using Scientific, Rockford, IL) according to the protocol of the manufacturer. A suspension the physicochemical property predictors of the Marvin calculator plugins. The of HLM (0.3 ml), with or without the TKI (in 1% DMSO), was placed in the “donor” values of fumic were predicted using the SimCYP ADME calculator (SimCYP side, and PB (0.5 ml) and DMSO were placed in the “receiver” side. Following prediction tools) based on the assigned compound type (acid/base/neutral) and at ASPET Journals on September 28, 2021 incubation, samples were treated as described below and analyzed by high- calculated log P and pKa and using equations based on log P given by Austin et al. performance liquid chromatography (HPLC) (Supplemental Table 1). (2002) (viz. eq. 11) and Hallifax and Houston (2006).

TABLE 1 Nonspecific binding of validation set drugs and TKIs to HLM (0.25 mg/ml)

fu 6 S.D. + CHAPS Ionized at Predicted fu Predicted fu by Predicted fu by Drug mic fu 6 S.D. 2 CHAPS Log P mic mic mic (6 mM) mic pH 7.4 by SimCYP Austin et al. (2002)a Hallifax and Houston (2006)b

% Validation set drugs Amitriptyline 0.91 6 0.03 0.64 6 0.07 99 4.81c–5.10d 0.26–0.35 0.13–0.17 0.25–0.35 Chlorpromazine 0.72 6 0.05 0.36 6 0.06 98 4.54c–5.20e 0.26–0.45 0.11–0.23 0.21–0.47 Felodipine 0.26 6 0.02 0.25 6 0.06 1 3.44c–4.83e 0.39–0.77 0.17–0.55 0.35–0.82 Isradipine 0.75 6 0.02 0.72 6 0.04 0 2.00c–4.30f 0.55–0.95 0.29–0.89 0.56–0.95 Loratidine 0.47 6 0.05 0.44 6 0.06 1 4.21d–5.94c 0.15–0.56 0.05–0.31 0.06–0.60 Midazolam 0.83 6 0.03 0.86 6 0.06 13 2.50f–3.97c 0.64–0.91 0.38–0.80 0.68–0.93 Nifedipine 0.88 6 0.03 0.87 6 0.02 1 1.82c–2.97e 0.85–0.96 0.69–0.91 0.89–0.96 Pazopanib 0.73 6 0.03 0.75 6 0.03 1 1.98e–3.59d 0.74–0.95 0.50–0.89 0.78–0.95 Zidovudine 1.00 6 0.02 1.01 6 0.04 100 0.53e–1.16c 0.98–0.99 0.96–0.98 0.97–0.98 TKIs Axitinib 0.79 6 0.03 0.2 4.15c–4.20f 0.58–0.59 0.31–0.33 0.60–0.62 Dabrafenib 0.81 6 0.06 0.79 6 0.02 64 3.54e–5.46c 0.33–0.75 0.08–0.52 0.14–0.79 Erlotinib 0.92 6 0.02 0.89 6 0.03 1 2.39e–3.30f 0.80–0.92 0.59–0.83 0.84–0.93 Gefitinib 0.76 6 0.05 22 3.75c–4.11e 0.70–0.60 0.34–0.45 0.63–0.74 Ibrutinib 0.57 6 0.04 13 2.76d–3.63c 0.73–0.89 0.49–0.75 0.77–0.91 Lapatinib 0.25 6 0.02 31 4.64c–5.18d 0.30–0.45 0.11–0.21 0.22–0.42 Nintedanib 0.93 6 0.03 59 2.59e–3.70d 0.72–0.91 0.47–0.78 0.76–0.92 Nilotinib 0.43 6 0.04 7 4.51d–5.36c 0.25–0.49 0.09–0.23 0.17–0.48 Regorafenib 0.14 6 0.03 0.1 4.20f–5.26e 0.27–0.58 0.10–0.31 0.19–0.60 Sorafenib 0.14 6 0.03 0.1 4.20f–5.16e 0.30–0.58 0.12–0.31 0.23–0.60 Trametinib 0.58 6 0.02 0 2.68e–3.45d 0.77–0.89 0.55–0.76 0.81–0.91

a 0.56 log P 2 1.41 Predicted values calculated using the equation fuinc =1/C · 10 +1. b 0.072(log P)2 + 0.067 × log P 2 1.126 Predicted values calculated using the equation fuinc = 1/1 + C · 10 . cMarvin. dALOGPS2.1. eACD/Laboratories. fXlogP3. 1936 Burns et al.

Results and Discussion cells were acceptable (.70%) and attainment of equilibrium (fumic – As noted in the Introduction, TKIs are typically lipophilic organic values of 0.85 1.05) was demonstrated for buffer-buffer and HLM- bases that are neutral or partially ionized at pH 7.4. The aqueous HLM control experiments. solubility of TKIs is poor, although, at least at relatively low concen- Based on these data, the microsomal binding of eight TKIs (axitinib, trations (,100 mM), solubilization is achieved in in vitro incubations erlotinib, gefitinib, ibrutinib, nilotinib, regorafenib, sorafenib, and containing HLM and DMSO (1% w/v). Initial studies of the NSB of trametinib) that are completely or largely unionized at pH 7.4 and – lapatinib, pazopanib, regorafenib, and sorafenib to HLM, using either three TKIs that are partially ionized (39 64%) at pH 7.4 (dabrafenib, equilibrium dialysis cells or a commercial RED device, demonstrated lapatinib, and nintedanib) was characterized in the presence of CHAPS that equilibrium was established in buffer-buffer and HLM-HLM (6 mM) (Table 1). NSB of the TKIs was concentration independent over – m (0.25 mg/ml) controls only for pazopanib. Increasing the concentration the range of 0.1 2 M. Measured fumic values ranged from 0.14 of DMSO present in the dialysis medium to 2% or using other organic (regorafenib and sorafenib) to 0.93 (nintedanib). Recoveries of each of . solvents (acetonitrile, methanol, and 0.95% acetonitrile plus 0.05% the TKIs from equilibrium dialysis cells were 70%, and attainment of equilibrium was demonstrated in buffer-buffer and HLM-HLM control DMSO, as reported by Kenny et al. (2012)] did not result in the at- experiments (fu values of 0.85–1.05). Subsequent experiments tainment of equilibrium in control experiments with lapatinib, regor- mic demonstrated that equilibrium was also attained in dialysis experiments afinib, and sorafenib. The mean (6 S.D.) fu of pazopanib obtained by mic with the neutral TKI erlotinib and the partially charged dabrafenib in the conventional equilibrium dialysis (in the presence of 1% DMSO) was absence of CHAPS; fumic values generated in the presence and absence 0.75 6 0.03 for five concentrations in the range of 0.1–2 mM(Table1). Downloaded from of detergent were essentially identical (Table 1). Similarly, the use of ultracentrifugation with lapatinib, regorafinib, and There was no clear relationship between the extent of ionization at sorafenib resulted in almost immeasurable concentrations in the pH 7.4 and fu for the drugs investigated here. While the two most reservoir compartment and low recoveries (,50%). mic extensively charged compounds (dabrafenib and nintedanib) exhibited Thus, we explored the use of the detergents Brij-58, CHAPS, and low NSB, the moderately ionized (38%) lapatinib bound extensively to Triton X-100 to aid the solubilization of TKIs in NSB experiments. Each HLM. Although the three highest binding TKIs (lapatinib, regorafenib, of the detergents used differ in terms of physicochemical properties and dmd.aspetjournals.org and sorafenib) have high log P values (4.2–5.2), the binding of axitinib critical micelle concentration (CMC). Since detergents may disrupt the (log P of 4.2) was comparatively low. Thus, log P is clearly not the sole microsomal membrane and hence potentially NSB, equilibrium dialysis determinant of extensive NSB of TKIs to HLM. We and others have experiments were performed with eight validation set basic drugs, for shown previously that NSB is generally highest for lipophilic organic which attainment of equilibrium could be demonstrated in buffer-buffer bases with molecular masses .200 Da (Austin et al., 2002; Hallifax and HLM-HLM (0.25 mg/ml) controls in the absence of detergent. and Houston, 2006; Sykes et al., 2006; McLure et al., 2011). Pazopanib (see above), felodipine, isradipine, loratidine, midazolam, and However, other physicochemical properties also influence the degree nifedipine are weak organic bases that are nonionized in phosphate buffer of NSB. These include the presence of halogen atoms (especially at ASPET Journals on September 28, 2021 at pH 7.4, whereas amitriptyline and chlorpromazine are lipophilic bases the trifluoromethyl group), polar surface area, charge distribution, that are essentially completely ionized at pH 7.4. Zidovudine, a zwitter- and number of hydrogen bond acceptors/donors [summarized by ionic compound that is extensively ionized and known not to bind McLure et al. (2011)]. Not only does the addition of fluorine or the “ ” nonspecifically to HLM (Uchaipichat et al., 2006), served as a negative trifluoromethyl groups to an aryl ring or adjacent to a heteroatom- control. Table 1 lists the percentage of ionization and the calculated log P containing functional group increase lipophilicity (Hagmann, 2008), for each of the control drugs and TKIs investigated here, whereas but halogenation of drugs is more generally known to enhance drug Supplemental Fig. 1 shows the structures of the validation set drugs and membrane binding and permeation. The importance of the trifluor- TKIs. It is noteworthy that calculated log P varied by as much as a factor omethyl group in particular has been highlighted in this regard of 2 (i.e., 2 log orders) using the four algorithms employed here (Table 1). (Gerebtzoff et al., 2004). With respect to the TKIs investigated here, Thus, Table 1 gives the range (lowest to highest) of the calculated log nilotinib, regorafenib, and sorafenib all possess a trifluoromethyl Pvalues. group (Supplemental Fig. 1). However, the two TKIs that bind most NSB of the validation set drugs amitriptyline, chlorpromazine, extensively to HLM, sorafenib and regorafenib, additionally contain felodipine, isradipine, loratidine, midazolam, nifedipine, and zidovu- one and two additional halogen atoms, respectively. dine was measured in duplicate at five concentrations in the range of Kenny et al. (2012) reported the NSB of a number of TKIs to HLM 1–100 mM, whereas the binding of all TKIs (including pazopanib) was (0.6 mg/ml) using a commercial RED device. The TKIs investigated measured at five concentrations in the range of 0.1–2 mM, which we included several of those studied here: lapatinib, nilotinib, pazopanib, have found to be relevant for inhibition studies (unpublished data). and sorafenib. As noted previously, in our hands, only pazopanib Binding of each of the validation set drugs was independent of attained equilibrium in HLM-HLM and buffer-buffer controls using a concentration in the absence of detergent. Brij-58 (0.007 mM; CMC, RED device and conventional equilibrium dialysis. The previous study – 0.007 0.07 mM) disrupted the binding of all eight basic drugs; fumic did not report data for HLM-HLM and buffer-buffer controls. Since values were higher with added detergent compared with the absence different HLM concentrations were employed in the two studies, the of detergent (data not shown). Despite the attainment of equilibrium in results are not directly comparable. However, the rank order of binding buffer-buffer and HLM-HLM controls in NSB experiments conducted reported by Kenny et al. (lapatinib . nilotinib . pazopanib . sorafenib) in the presence of Triton X-100 (CMC, 0.2–0.9 mM), the recoveries of differs markedly from that found here. each compound were low (,50%). By contrast, the presence The SimCYP ADME calculator and equations published by Austin of CHAPS (6 mM; CMC, 6–10 mM) had no effect on the binding of et al. (2002) and Hallifax and Houston (2006) that were used here to pazopanib, felodipine, isradipine, loratidine, midazolam, and nifedipine predict fumic values take into account only log P and charge state/ to HLM, although binding of the fully charged bases amitriptyline and ionization. The predicted fumic values are shown in Table 1 as a range chlorpromazine to HLM was disrupted (Table 1). No binding of since the calculated log P values differed among the algorithms. Indeed, zidovudine was observed in the absence or presence of CHAPS. as highlighted previously, the calculated log P values surprisingly Importantly, recoveries of all nine drugs from the equilibrium dialysis differed by as much as a factor of 2. Despite using a range of log P The Nonspecific Binding of Tyrosine Kinase Inhibitors to HLM 1937 values for the calculation of fumic, this parameter was generally Department of Clinical KUSHARI BURNS underpredicted for the most highly bound compounds using the Pharmacology and Flinders Centre PRAMOD C. NAIR SimCYP calculator and the equation published by Hallifax and for Innovation in Cancer, School of ANDREW ROWLAND Houston. Although the fumic values predicted for the highly bound Medicine, Flinders University, PETER I. MACKENZIE compounds regrofenib and sorafenib using the equation of Austin et al. Adelaide, Australia KATHLEEN M. KNIGHTS spanned the experimentally determined fumic, this equation tended to JOHN O. MINERS overpredict the binding of other TKIs. No single algorithm for calculated log P performed consistently well for predicting fumic (data not shown). More generally, the choice of log P algorithm used for the Authorship Contributions prediction of NSB and other parameters warrants further investigation. Participated in research design: Miners, Burns, Knights, Mackenzie. Available evidence demonstrates that NSB, particularly of lipophilic Conducted experiments: Burns, Rowland. basic drugs, arises largely from the partitioning of drugs into Performed data analysis: Miners, Burns, Nair, Rowland. phospholipid membranes (Margolis and Obach, 2003; Nussio et al., Wrote or contributed to writing of the manuscript: Miners, Knights, Rowland, 2007; McLure et al., 2011; Nagar and Korzekwa, 2012). Neutron Nair, Burns. diffraction studies with the charged lipophilic base propranolol showed that that the lipophilic naphthalene moiety partitions into the lipid References bilayer, whereas the charged amine group apparently associates with

Austin RP, Barton P, Cockroft SL, Wenlock MC, and Riley RJ (2002) The influence of non- Downloaded from the phospholipid head group (Herbette et al., 1983). CHAPS was specific microsomal binding on apparent intrinsic clearance, and its prediction from physico- selected as the detergent for the solubilization of TKIs here because it chemical properties. Drug Metab Dispos 30:1497–1503. Bowalgaha K, Elliot DJ, Mackenzie PI, Knights KM, Swedmark S, and Miners JO (2005) retained all compounds in an aqueous solution and, since the S-Naproxen and desmethylnaproxen glucuronidation by human liver microsomes and concentration employed in this work (6 mM) is at the lower end of recombinant human UDP-glucuronosyltransferases (UGT): role of UGT2B7 in the elimi- nation of naproxen. Br J Clin Pharmacol 60:423–433. the CMC range, it was expected not to affect dialysis. This was Gerebtzoff G, Li-Blatter X, Fischer H, Frentzel A, and Seelig A (2004) Halogenation of drugs confirmed by the attainment of equilibrium in control experiments. enhances membrane binding and permeation. ChemBioChem 5:676–684.

Grime K and Riley RJ (2006) The impact of in vitro binding on in vitro-in vivo extrapolations, dmd.aspetjournals.org Moreover, CHAPS (6 mM) was shown to have no effect on the NSB projections of metabolic clearance and clinical drug-drug interactions. Curr Drug Metab 7:251–264. of the six neutral basic validation set drugs: felodipine, isradipine, Hagmann WK (2008) The many roles for fluorine in medicinal chemistry. J Med Chem 51: 4359–4369. loratidine, midazolam, nifedipine, and pazopanib. Later experiments Hallifax D and Houston JB (2006) Binding of drugs to hepatic microsomes: comment and further demonstrated that CHAPS (6 mM) is similarly without effect on assessment of current prediction methodology with recommendation for improvement. Drug Metab Dispos 34:724–726. the NSB of erlotinib and the partially charged dabrafenib (Table 1). Herbette L, Katz AM, and Sturtevant JM (1983) Comparisons of the interaction of propranolol CHAPS concentrations below 6 mM (viz. 2 and 4 mM) appeared to give and timolol with model and biological membrane systems. Mol Pharmacol 24:259–269. Houston JB (1994) Utility of in vitro drug metabolism data in predicting in vivo metabolic incomplete solubilization of the more lipophilic TKIs, whereas fumic clearance. Biochem Pharmacol 47:1469–1479. values of representative TKIs and “control” drugs (felodipine, lapatinib, Kalipatnapu S and Chattopadhyay A (2005) Membrane protein solubilization: recent advances at ASPET Journals on September 28, 2021 and challenges in solubilization of serotonin1A receptors. IUBMB Life 57:505–512. pazopanib, regorafenib, and sorafenib) were shown to increase as the Kenny JR, Mukadam S, Zhang C, Tay S, Collins C, Galetin A, and Khojasteh SC (2012) Drug- CHAPS concentration was increased to 10 mM (data not shown). The drug interaction potential of marketed oncology drugs: in vitro assessment of time-dependent latter observation appears to be consistent with the known effects of cytochrome P450 inhibition, reactive metabolite formation and drug-drug interaction pre- diction. Pharm Res 29:1960–1976. CHAPS on membrane structure. At low concentrations, detergent Margolis JM and Obach RS (2003) Impact of nonspecific binding to microsomes and phos- molecules tend only bind to the surface of the membrane, but as the pholipid on the inhibition of cytochrome P4502D6: implications for relating in vitro inhibition data to in vivo drug interactions. Drug Metab Dispos 31:606–611. CHAPS concentration increases, the membrane bilayer becomes McLure JA, Birkett DJ, Elliot DJ, Williams JA, Rowland A, and Miners JO (2011) Application of disrupted (Kalipatnapu and Chattopadhyay, 2005). Such a model the fluorescent probe 1-anilinonaphthalene-8-sulfonate to the measurement of the nonspecific binding of drugs to human liver microsomes. Drug Metab Dispos 39:1711–1717. further explains why CHAPS disrupts the binding of the fully charged McLure JA, Miners JO, and Birkett DJ (2000) Nonspecific binding of drugs to human liver amitriptyline and chlorpromazine to HLM. microsomes. Br J Clin Pharmacol 49:453–461. Miners JO, Knights KM, Houston JB, and Mackenzie PI (2006) In vitro-in vivo correlation for In summary, conventional equilibrium dialysis performed in the drugs and other compounds eliminated by glucuronidation in humans: pitfalls and promises. presence of CHAPS permits measurement of the NSB of neutral TKIs Biochem Pharmacol 71:1531–1539. Miners JO, Mackenzie PI, and Knights KM (2010) The prediction of drug-glucuronidation pa- to HLM. NSB data were also generated for partially charged TKIs, rameters in humans: UDP-glucuronosyltransferase enzyme-selective substrate and inhibitor which appeared consistent with the physicochemical properties of these probes for reaction phenotyping and in vitro-in vivo extrapolation of drug clearance and drug- drug interaction potential. Drug Metab Rev 42:196–208. compounds (and confirmed in the case of dabrafenib). It is emphasized Nagar S and Korzekwa K (2012) Commentary: nonspecific protein binding versus membrane that the method has been developed specifically for uncharged and partitioning: it is not just semantics. Drug Metab Dispos 40:1649–1652. Nussio MR, Sykes MJ, Miners JO, and Shapter JG (2007) Characterisation of the binding of weakly charged lipophilic organic bases, and it is unlikely to be cationic amphiphilic drugs to phospholipid bilayers using surface plasmon resonance. Chem- applicable to other chemical classes. Moreover, experiments were MedChem 2:366–373. performed at a single HLM concentration (0.25 mg/ml) and further Obach RS (1999) Prediction of human clearance of twenty-nine drugs from hepatic microsomal intrinsic clearance data: An examination of in vitro half-life approach and nonspecific binding method optimization will be required for other HLM concentrations. to microsomes. Drug Metab Dispos 27:1350–1359. Since detergents are known to affect cytochrome P450 and UDP- Reyner EL, Sevidal S, West MA, Clouser-Roche A, Freiwald S, Fenner K, Ullah M, Lee CA, and Smith BJ (2013) In vitro characterization of axitinib interactions with human efflux and glucuronosyltransferase enzyme activities in a concentration-dependent hepatic uptake transporters: implications for disposition and drug interactions. Drug Metab and variable manner, the addition of CHAPS to microsomal incubations Dispos 41:1575–1583. Rostami-Hodjegan A and Tucker GT (2007) Simulation and prediction of in vivo drug metab- is generally not recommended. However, as noted earlier, the solubility olism in human populations from in vitro data. Nat Rev Drug Discov 6:140–148. of relatively low concentrations of TKIs in incubations of HLM Sykes MJ, Sorich MJ, and Miners JO (2006) Molecular modeling approaches for the prediction of the nonspecific binding of drugs to hepatic microsomes. J Chem Inf Model 46:2661–2673. containing DMSO appears to be acceptable. NSB has generally been Uchaipichat V, Winner LK, Mackenzie PI, Elliot DJ, Williams JA, and Miners JO (2006) Quanti- ignored in previous in vitro kinetic and inhibition studies of TKIs. The tative prediction of in vivo inhibitory interactions involving glucuronidated drugs from in vitro – substantial microsomal binding of many of the TKIs investigated here data: the effect of fluconazole on zidovudine glucuronidation. Br J Clin Pharmacol 61:427 439. indicates that the K and K values will be significantly overestimated if m i Address correspondence to: Professor J.O. Miners, Department of Clinical NSB is not accounted for, which in turn impacts the accuracy of in vivo Pharmacology, Flinders University School of Medicine, GPO Box 2100, Adelaide, intrinsic clearance values and the drug-drug interaction potential SA 5001, Australia. E-mail [email protected] predicted using in vitro–in vivo extrapolation approaches. Kushari Burns, Pramod C Nair, Andrew Rowland, Peter I Mackenzie, Kathleen M Knights and John O Miners. The non-specific binding of tyrosine kinase inhibitors to human liver microsomes. Drug Metabolism and Disposition.

Supplemental data: Supplemental Table 1 and Supplemental Figure 1

Supplemental Table 1: HPLC conditions for validation set drugs and TKIs

Drug Mobile phase Detector Retention time composition wavelength (min) (nm)

Amitriptyline 37.5%B : 62.5%C 240 5.9 Axitinib 85%B : 15%C 350 6.9 Chlorpromazine 40 %B : 60%C 254 2.7 Dabrafenib 50%A : 50%C 331 3.6 Erlotinib 60%A : 40%C 260 3.4 Felodipine 40%A : 60%C 235 3.3 Gefitinib 40%A : 60%C 330 2.6 Ibrutinib 50%A : 50%C 254 3.2 Isradipine 40%D : 60%C 325 2.4 Loratidine 60%B : 40%C 254 2.6 Midazolam 45%A : 55%C 254 3.8 Lapatinib 60%A : 40%C 265 3.7 Nifedipine 60%A : 40%C 235 3.3 Nilotinib 50%A : 50%C 250 3.7 Nintedanib 85%B : 15%C 360 4.8 Pazopanib 70%A : 30%C 265 4.1 Regorafenib 35%A : 65%C 265 2.9 Sorafenib 50%A : 50%C 265 2.8 Trametinib 50%A : 50%C 265 3.3 Zidovudine 92%B : 8%C 267 2.5 Mobile phase composition: A: 10 mM ammonium acetate (adjusted to pH 5.7 with glacial acetic acid) containing 10% acetronitrile. B: 10 mM triethylamine (adjusted to pH 2.5 with 12 M perchloric acid) containing 10% acetonitrile C: acetonitrile D: Water containing 5% acetonitrile

Supplemental Figure 1. Chemical structures of tyrosine-kinase inhibitors (TKIs) and validation set drugs.

Tyrosine-kinase inhibitors (TKIs)

Axitinib Dabrafenib

Erlotinib Gefitinib

Ibrutinib Lapatinib

Nilotinib Nintedanib

Pazopanib Regorafenib

Sorafenib Trametinib

Validation set drugs

Amitriptyline Chlorpromazine Felodipine

Isradipine Loratadine Midazolam

Nifedipine Zidovudine